In this paper, RasID is enhanced and evaluated systematically. RasID is an abbreviation of Random Search with Intensification and Diversification. RasID can search for a global minimum based on a probabilitiy density function, which can be modified adaptively using information based on success and failure of the past searching. As a result, RasID performs intensified search and diversified search iteratively to find a global minimum.
The improvements of RasID are, (1) the probability density function is modified so that each variable can have its own unique density function, and (2) multiple candidates for solutions are created in the search to enhance the efficiency of the searching.
Modified RasID is compared systematically with typical optimization methods such as Evolutionary. Programming and Fast Evolutionary Programming introduced recently using 23 different complibated functions. From simulation results, it has been clarified that the performance of Modified RasID is comparable to the ones of EP and FEP in spite of using one individual.